Professional Documents
Culture Documents
Mobile Cloud Computing On Education
Mobile Cloud Computing On Education
A hybrid modeling approach for predicting the educational use of mobile cloud
computing services in higher education
Ibrahim Arpaci
PII: S0747-5632(18)30445-X
DOI: 10.1016/j.chb.2018.09.005
Please cite this article as: Ibrahim Arpaci, A hybrid modeling approach for predicting the
educational use of mobile cloud computing services in higher education, Computers in Human
Behavior (2018), doi: 10.1016/j.chb.2018.09.005
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form.
Please note that during the production process errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
A hybrid modeling approach for predicting the educational use of mobile cloud computing
ABSTRACT
The decision to integrate mobile cloud computing (MCC) in education without determining optimal
use scenarios is a universal problem as the adoption of such services becomes widespread.
Accordingly, this study developed and validated a predictive model that explains the role of
students’ information management (i.e. retrieve, store, share, and apply) practices in predicting
their attitudes toward using the MCC services for educational purposes. This study validated the
model by the complementary use of machine learning algorithms alongside a classical SEM-based
approach based on data collected from 308 undergraduate students. The SEM results indicated that
the students’ information management (i.e. retrieve, store, share, and apply) practices were
significantly associated with their attitudes, which were significantly associated with the behavioral
intentions. The structural model explained a significant portion of the variance in the behavioral
intentions. Likewise, the classifier model suggested that the students’ information management
practices and attitudes predicted the behavioral intentions. Further, the applied algorithms predicted
the behavioral intentions with an accuracy of more than 72% in most cases. Thereby, the study
extended an original theory (TRA) into the MCC area by using a multi-analytical approach. The
findings implied that employing the MCC services for personal information management should
be supported and encouraged in the higher education by designing authentic learning environments
ACCEPTED MANUSCRIPT
A hybrid modeling approach for predicting the educational use of mobile cloud computing
ABSTRACT
The decision to integrate mobile cloud computing (MCC) in education without determining optimal
use scenarios is a universal problem as the adoption of such services becomes widespread.
Accordingly, this study developed and validated a predictive model that explains the role of
students’ information management (i.e. retrieve, store, share, and apply) practices in predicting
their attitudes toward using the MCC services for educational purposes. This study validated the
model by the complementary use of machine learning algorithms alongside a classical SEM-based
approach based on data collected from 308 undergraduate students. The SEM results indicated that
the students’ information management (i.e. retrieve, store, share, and apply) practices were
significantly associated with their attitudes, which were significantly associated with the behavioral
intentions. The structural model explained a significant portion of the variance in the behavioral
intentions. Likewise, the classifier model suggested that the students’ information management
practices and attitudes predicted the behavioral intentions. Further, the applied algorithms predicted
the behavioral intentions with an accuracy of more than 72% in most cases. Thereby, the study
extended an original theory (TRA) into the MCC area by using a multi-analytical approach. The
findings implied that employing the MCC services for personal information management should
be supported and encouraged in the higher education by designing authentic learning environments
1. Introduction
Mobile cloud computing (MCC) as a new distributed computing paradigm can be defined as an
infrastructure, application or process, where the data storage and processing migrated from smart
mobile technologies to the distributed cloud servers (Dinh, Lee, Niyato, & Wang, 2013). The MCC
has gained a substantial attention of both the organizations and individuals as a promising solution
for the ubiquitous environments, in which data storage and processing occur over a “cloud” via the
Internet (Park & Kim, 2014). Investing in the MCC, organizations improved their capacity and
capabilities without the cost of installing new infrastructure or software (Subashini & Kavitha,
2011). Further, the MCC reduced the initial installation and maintenance costs, and thereby,
There are various cloud deployment-models such as private, public, hybrid, and community cloud
(Shon, Cho, Han, & Choi, 2014). The study focused on the public cloud that allocates resources on
a per-user basis through applications such as Dropbox, SkyDrive, and Google Drive. However, the
MCC service models for individuals include “platform as a service” (PaaS) for automatic content
synchronizing and “infrastructure as a service” (IaaS) for information storage and management.
On the other hand, these services provide enterprises with the IaaS for an enterprise infrastructure
building and “software as a service” (SaaS) for an enterprise solution provider (Arpaci, 2016).
The MCC provides users at the organizational and individual level with several advantages and
given these advantages the MCC services are considered to be the growth engine of the industry
4.0 (Park & Kim, 2014). For example, the MCC increases data storage capacity and allows an
efficient data synchronization and information management in a ubiquitous environment (Park &
Kim, 2014). Besides, energy efficiency is a critical issue since battery power of the mobile
technologies is limited. Migration of the complex processing from mobile technologies to the cloud
ACCEPTED MANUSCRIPT
3
servers extends the battery life (Cuervo et al., 2010). Furthermore, service providers have certain
security mechanisms and backup systems to protect user data. Therefore, saving documents and
files on the distributed cloud servers may enhance the reliability by reducing virus-like threats and
Smart mobile technologies change the way of information is managed (Ogiela, 2017). For example,
using mobile Internet on smartphones provides students with ubiquitous access to information and
learning materials. Further, social networking and communication applications allow them to share
information with peers. More importantly, the integrated technologies enable them to use
information stored in the MCC services in problem solving and decision-making. Considering the
extensive use of mobile devices such as smartphones, tablets, and smartwatches by undergraduate
students, it is important to find creative ways for an effective integration of these technologies into
the higher education. Such an integration may enable students to learn by using the technologies,
which they are familiar and confident with, and highly motivated to use. It is important to note that
storage, sharing, and application is one of the main advantages of the MCC services. However, the
decision to integrate the MCC in educational settings without determining optimal use scenarios is
a universal problem as the adoption of such services becomes widespread. Thus, this study
investigated adoption of the MCC among undergraduate students by focusing on the use of these
2. Literature Review
The MCC is a recent research area and there are numerous studies on the adoption of the MCC at
the both organizational and individual level. For example, Ul Amin, Inayat, Shahzad, Saleem, and
Aijun (2017) applied the “Unified Theory of Acceptance and Use of Technology” (UTAUT) to
ACCEPTED MANUSCRIPT
4
identify key predictors of the MCC adoption by healthcare professionals. Their results suggested
significant in determining the behavioral intentions. In another study, Lian, Yen, and Wang (2014)
investigated the factors affecting the cloud computing adoption in Taiwan’s hospital industry based
Environment) framework. The results suggested that technical competence, data security, cost,
complexity, and top management support were critical factors in the adoption. Chen, Chen, and
Lee (2018) investigated key factors affecting the organizations’ adoption of the cloud services.
Their results suggested that top-management-support has a vital role in the adoption at the
organizational level.
Lee (2016) identified the key factors explaining the adoption of cloud services at the individual
level based on the “Diffusion of Innovations” (DOI) theory. The results suggested the relative
advantage, observability, and self-efficacy were significant antecedents of the adoption. Park and
Kim (2014) focused on key factors predicting the individuals’ acceptance of the MCC services.
Results indicated that connectedness, perceived mobility, service quality, security, and satisfaction
were significant predictors of the user acceptance. Wang and Huang (2016) reported that social
influence was a key predictor of the students’ intention to use cloud services. Further, they reported
that scaffolding in problem-solving and training were useful in familiarizing the students with such
services.
Arpaci (2016) proposed a theoretical structural model based on the “Technology Acceptance
Model” (TAM) and suggested that subjective norm, trust, and perceived-usefulness play significant
roles in predicting the students’ attitudes towards using the MCC services. In another study,
Sharma, Al-Badi, Govindaluri, and Al-Kharusi (2016) developed a hybrid model based on the
ACCEPTED MANUSCRIPT
5
TAM to predict factors affecting the adoption of the cloud computing by IT professionals. They
employed linear regression along with neural network analysis and found that job opportunity was
the most significant predictor of the adoption. Shiau and Chau (2016) developed a theoretical
structural model based on the TAM and motivational model to predict behavioral intentions to join
a cloud computing class. The results suggested subjective norms and attitudes were positively
associated with the behavioral intention. Kim and Kim (2018) proposed a research model in order
to identify the antecedents of the MCC adoption. The results suggested that trust, convenience, and
3. Theoretical Background
Most of the studies reviewed applied or extended a theory such as TAM (Davis, 1989) or UTAUT
(Venkatesh, Morris, Davis, & Davis, 2003) by conducting an explanatory statistical analysis.
However, such an approach has two main limitations. First, each theory was developed and tested
for a specific domain, and therefore, may not work well in other domains or areas. Thus, this study
extended the original “Theory of Reasoned Action” (TRA) with additional constructs to better
explain the research context. Second, previous studies pointed out statistical limitations of the
approach by combining predictive analytics and causal explanatory statistical analysis to validate
a predictive model (Sharma et al., 2016; Sharma, Joshi, & Sharma, 2016; Tan, Ooi, Leong, & Lin,
2014). The study therefore tested the research model by using both the classical SEM approach and
artificial intelligence techniques (i.e. machine learning). The study employed the SEM approach to
understand causal relationships and the (complementary) study applied the machine learning
algorithms to predict the behavioral intentions based on the proposed constructs. Further, this study
approaches the MCC adoption from an information management perspective. The fact that
ACCEPTED MANUSCRIPT
6
providing an effective information management was one of the main advantages of the MCC
services supports the idea that individuals’ information management practices may have a critical
role in their adoption decision. Accordingly, this study investigated the role of students’
information management practices on their attitudes toward using the MCC services in educational
settings.
The TRA postulated that beliefs affect attitudes, which positively influence behavioral intentions,
whereas the intentions ultimately influence the actual behavior (Fishbein & Ajzen, 1975). Benbasat
and Barki (2007) suggested using the original theory (TRA) to provide a strong theoretical
grounding for a novel model by incorporating different antecedents relevant to the nature of ICT
integration and use in diverse settings. Likewise, Nistor (2014) argued that “educational technology
acceptance” studies should consider characteristics of the educational context. Considering the
shortcomings of the technology acceptance models and analysis methods suggested by (Benbasat
& Barki, 2007), this study identified the key antecedents of the attitudes towards using the MCC
services and employed an innovative approach comprising both machine learning algorithms and
the SEM, and thereby, strengthened the study design. Figure 1 illustrates the research model that
suggests the behavioral intentions (continued use intentions) are predicted by the attitudes, which
are predicted by the students’ information management practices such as information retrieval,
Rogers’s (2003, p.19) DOI theory suggested that the stages by which an individual adopts a new
technology or innovation, whereby diffusion is accomplished include “awareness of the need for
an innovation, decision to adopt (or reject) the innovation, initial use of the innovation to test it,
and continued use of the innovation.” Accordingly, a training program introduced the participants
performs to create or acquire, store, retrieve, share, and apply the information required to complete
the tasks. The participants employed the mobile web services (i.e. search, maps), mobile cloud
applications (i.e. Google Drive, iCloud, Dropbox) and social media applications to perform the
training tasks.
3.1. Hypotheses
“Behavioral intention” was defined as “the degree of an individual’s belief that he or she will
continue to use a system” (Arpaci, 2017, p. 384). However, attitudes towards using a technology
was defined as “an individual’s overall affective reaction to use a particular system” (Arpaci, 2017,
p. 384). This study therefore hypothesized that the attitude toward using the MCC services would
be positively associated with the behavioral intention to use such services for educational purposes.
H1a. Attitude towards using the MCC services would be positively associated with the behavioral
intention.
H1b. Attitude towards using the MCC services would predict the behavioral intention.
Information is processed, categorized, organized or structured data (Nonaka, 1994; Davenport &
insights (Nonaka & Takeuchi, 1995). Knowledge includes organizational routines, values, and
practices; therefore, this study preferred to use the term “information management” instead of
“knowledge management.” This study defined information management as practices that help
retrieve and store information, share it with others, and apply it to solve problems and make better
ACCEPTED MANUSCRIPT
8
and timely decisions (Alavi & Leidner, 2001; Mitchell, 2003). Turban, Sharda, and Delen (2011)
retrieval and access, transfer, storage, and application. In the same vein, the present study suggested
the integration of smart mobile technologies into the higher education to support personal
information management. The more information managed by using the MCC services the more
favorable the attitude towards using such services for educational purposes. Further, the students’
information management practices would be a predictor of the behavioral intention to use the MCC
H2a. An increase in the volume of information that’s retrieved via the use of the MCC services is
positively associated with the attitude towards the use of such services.
H2b. Information retrieval would predict the behavioral intention to use the MCC services for
educational purposes.
H3a. An increase in the volume of information that’s stored via the use of the MCC services is
positively associated with the attitude towards the use of such services.
H3b. Information storage would predict the behavioral intention to use the MCC services for
educational purposes.
H4a. An increase in the volume of information that’s shared via the use of the MCC services is
positively associated with the attitude towards the use of such services.
H4b. Information sharing would predict the behavioral intention to use the MCC services for
educational purposes.
ACCEPTED MANUSCRIPT
9
H5a. An increase in the volume of information that’s applied via the use of the MCC services is
positively associated with the attitude towards the use of such services.
H5b. Information application would predict the behavioral intention to use the MCC services for
educational purposes.
4. Method
4.1. Sample
A sample of 308 university students were recruited from a public university in Turkey. The study
included 186 females (60.4%) and 122 (39.6%) males with a mean age of 21.88 years (SD=2.37).
Majority of the students (42.5%) were seniors; 12.3% were freshmen, 27.6% were sophomore, and
4.2. Procedure
IRB of the affiliated university approved the research and the procedures complied with the ethical
standards of the institutional board guidelines. All participants were informed about the purpose of
the research after an informed consent was obtained. The study was conducted in regularly
scheduled IT classes (4 hours per week) during 14 weeks. The participants performed practical
implementations of the theoretical knowledge during the interactive tasks such as: 1) Search for
research articles in electronic databases (i.e. Scopus, WOS, and Google Scholar), 2) Save an
annotated bibliography of the articles to the MCC services (i.e. OneDrive, Dropbox, and Google
Drive), 3) Form a communities of practice and use the MCC services along with social networks
to share files, communicate with others, and build a common cognition to complete the tasks.
Finally, the survey instrument with explicit assurance that anonymity is guaranteed was
4.3. Instruments
The measurement instrument for the variables “behavioral intentions” and “attitudes” was a Likert
scale adapted from Ajzen’s theory (1991). Moreover, the measurement instrument for the
information management practices of retrieval, storage, sharing, and application was adapted from
a previously validated and reliable instrument (Arpaci, 2017). The final instrument has 24 items
with a five-point Likert-type scale ranging from “strongly disagree” to “strongly agree.”
5. Results
IBM SPSS (ver. 25) and IBM AMOS (ver. 23) were used to screen and analyze the data. Construct
validity along with reliability of the scales were assessed by conducting an EFA (See Table 1).
Results revealed that each item had a communality value and a factor loading greater than the
critical threshold (.40) suggested by Field (2005). Cronbach’s alpha reliability analysis results
indicated a good homogeneity and reliability among the items. The AVE values were greater than
.50, indicating adequate convergent validity for all constructs (Hair, Tatham, Anderson, & Black,
2006).
Kaiser’s (1970) “KMO measure of sampling adequacy” and Bartlett’s (1951) “test of sphericity”
suggested that conducting a CFA on the data was appropriate (See Table 2). KMO values were all
above .65 and Bartlett’s test values were all significant (p< .001). This suggested that measures for
The CFA was conducted using maximum-likelihood estimates in an attempt to test the research
model. Results suggested the structural model demonstrates a “good fit” to data: [χ2/df=1.67,
GFI=.91, NFI=.92, AGFI=.88, CFI=.96, TLI=.96, IFI=.96, RMSEA=.05]. A GFI, NFI, TLI, CFI,
and IFI equal or above .90 and a RMSEA less than .08 indicate an acceptable model fit (Hair et al.,
2006). Further, Kline (2005) suggested a χ2/df less than 2.0 indicates a good fit.
A SEM approach was used to explore causal relationships among the variables. The standardized-
path-coefficients in the structural model were all statistically significant (See Figure 1). Results
indicated that information retrieval, information storage, information sharing, and information
application were significantly associated with the attitudes toward using the MCC services. All
these factors explained .33 of the variance in the attitudes. Whereas, the information retrieval had
the largest impact on the attitudes with a standardized coefficient of .45. On the other hand, the
information storage had a relatively small impact on the attitudes with a standardized coefficient
of .16.
The results shown in Table 3 suggested the attitudes were significantly associated with the
behavioral intentions with a standardized coefficient of .77. These factors accounted for a
5.4. Testing the Predictive Model using the Machine Learning Algorithms
Classification is one of the machine learning techniques that aims to build a classifier model in
order to predict a behavior through classifying the data into a number of pre-defined classes based
ACCEPTED MANUSCRIPT
12
on a certain criterion (Ngai, Xiu, & Chau, 2009). Common techniques employed for the
classification are Bayesian networks, decision trees, if-then-else rules, and neural networks.
searches for correlations between the attributes and performance of the learning algorithms
(Kotsiantis, Zaharakis, & Pintelas, 2007). This study employed machine learning classification
algorithms to predict the behavioral intentions based on attributes of the students’ information
management practices and their attitudes. Behavioral intentions and attitudes have nominal values
and classified as low, medium or high by using the Mean ± SD combination. On the other hand,
the attributes of information retrieval, storage, sharing, and application have numeric values.
The predictive model was tested by using Weka (ver. 3.8.1) based on the most common
classification algorithms, which classify data into a number of pre-defined classes with different
learning schemas. The results shown in Table 4 provided performance of the best performance
classifier based on two test modes; 10-fold cross-validation and percentage split (66.0%).
The study employed a linear logistic regression classifier (Logistic), a meta AdaBoostM1, a
Bayesian classifier (Naive bayes), a lazy LWL, and two rule learners (OneR, JRip). The results
showed that the OneR rule learner performs slightly better than the other classifiers. OneR classifier
predicted the behavioral intentions with an accuracy of 73.70% and 71.43% for 10-fold cross-
validation and percentage split, respectively. Further, this algorithm had a better performance in
terms of true positive (TP) rate (.74) and F-measure (.73) for the 10-fold test mode compared to
the other classifiers (i.e. Naive bayes, Logistic, JRip, AdaBoostM1, LWL). The results shown in
Table 5 suggested that attributes of the attitudes, information retrieval, information storage,
ACCEPTED MANUSCRIPT
13
information sharing, and information application predicted the behavioral intentions with an
accuracy of 74%, and thereby, H1b, H2b, H3b, H4b, and H5b were supported.
Using the MCC services provides students with ubiquitous access to instructional materials and
information anytime and anywhere. In addition, these services provide students with a storage
capacity to store their reference materials. Furthermore, the MCC services enable the students to
share ideas, documents, and files with peers and classmates. Eventually, the students can apply and
transform that knowledge and experience in problem-solving and decision-making. However, the
decision to integrate these services in educational settings requires determining an optimal use
scenario. This study therefore focused on the use of the MCC services for personal information
management. Accordingly, the study employed a hybrid modeling approach for investigating the
role of students’ information management (i.e. retrieve, store, share, and apply) practices in their
attitudes (toward use) and the intentions to (continue to) use the MCC services.
The actuality of the ideas and employing an integrated multi-analytical approach were the novel
contributions of the study. The study developed a research model by extending the TRA with
additional constructs and tested the model by integrating the SEM and machine learning approach.
This study identified the causal relationships between the dependent and independent variables by
employing the SEM approach. The SEM results indicated that the students’ information
management practices (i.e. retrieve, store, share, and apply) and attitudes were significantly
associated with the behavioral intentions to use the MCC services for educational purposes.
ACCEPTED MANUSCRIPT
14
In a complementary study, machine learning algorithms were employed to explore whether the
students’ information management practices and attitudes would predict the behavioral intentions.
The study tested a number of classification algorithms with different methodologies, including
bayesian networks, decision trees, if-then-else rules, association rules, and neural networks. The
results indicated that the OneR rule learner performs slightly better than other classifiers. It is
worthy to mention that OneR (or one class) algorithm works well if one of the attributes can predict
the outcome (dependent) variable better than the other attributes in the training data (Holte, 1993).
The classifier model suggested that the attitude predicts the behavioral intention better than the
other attributes (CCI=.74). In line with these results, the structural model suggested that the attitude
was a significant factor in predicting the behavioral intention (β=.77). Further, the SEM results
suggested the students’ information management practices and attitudes were significantly
associated with the behavioral intentions and accounted for 59% of the variance in the behavioral
intentions.
It is important to note that the relation between the attitudes and the behavioral intentions explains
a lot of variance in the model, which is desirable, but not informative or useful. Conceptually, the
attitudes (toward use) and the intentions to (continue to) use are a (conceptual) component of the
later. However, this study overcomes this design flaw by capitalizing on its two main strengths: 1)
the complementary use of parametric and nonparametric approaches, and 2) the extension of the
variables. In the model, both the attitudes and behavioral intentions were endogenous (dependent)
variables. The SEM results identified the causal relationships among the endogenous variables and
between the endogenous and exogenous variables. On the other hand, the classifier model included
ACCEPTED MANUSCRIPT
15
the attitudes along with information management practices as predictors of the behavioral
intentions. Although, presentation of the two analyses (SEM vs. machine learning) seems two
separate results, considering the operationalization of the same constructs in both model, indeed,
this study presents a “hybrid modelling” approach and not a mere dual-perspective approach.
performance and efficiency of the learning tasks. This implies that employing the MCC services
for personal information management should be supported and encouraged in the higher education
by designing authentic learning environments and scaffolding students in using such services.
Universities should develop their own policies (i.e. bring-your-own-device policy) and cloud-based
applications to maximize benefits of the such services. More importantly, instructors should
employ blended learning strategies that require information management practices to promote
The SEM does not provide a complete solution to all statistical problems of the prediction. On the
other hand, it provides a powerful means of testing a causal model based on a well-specified
conceptual theory. Accordingly, the complementary use of the SEM and the machine learning
algorithms to develop a prediction model provided unique information on the performance of the
SEM-based prediction compared to the other prediction models. Besides, using two approaches,
one focused on contrasting/comparing and another on completing, help make robust predictions.
However, the study has a number of limitations. This study did not account for the potential
(Arpaci, 2015; Arpaci, Cetin Yardimci, & Turetken, 2015). Therefore, one should be careful to
ACCEPTED MANUSCRIPT
16
generalize the findings to the samples with a different socio-cultural background. Concerns
reported by (Stergiou & Psannis, 2017) may also affect the degree of adoption. Future studies may
therefore focus on the concerns of confidentiality, privacy, latency, performance instability, data
security, and lack of service-level-agreements. The fact that an effective technology integration
into the higher education requires a techno-pedagogical approach and needs to consider the aspects
implies that future studies should take into account all these aspects. Finally, a longitudinal study
References
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision
Alavi, M., & Leidner, D. E. (2001). Review: Knowledge management and knowledge management
systems: Conceptual foundations and research issues. MIS Quarterly, 25(1), 107-136.
Arpaci, I. (2015). A comparative study of the effects of cultural differences on the adoption of
10.1111/bjet.12160
Arpaci, I. (2016). Understanding and predicting students’ intention to use mobile cloud storage
10.1016/j.chb.2017.01.024
Arpaci, I., Cetin Yardimci, Y., & Turetken, O. (2015). A cross-cultural analysis of smartphone
Benbasat, I., & Barki, H. (2007). Quo vadis TAM? Journal of the Association for Information
Chen, S. L., Chen, J. H., & Lee, Y. H. (2018). A comparison of competing models for
Cuervo, E., Balasubramanian, A., Cho, D. K., Wolman, A., Saroiu, S., Chandra, R., & Bahl, P.
(2010). MAUI: Making smartphones last longer with code offload. In Proceedings of the 8th
international conference on Mobile systems, applications, and services (pp. 49-62). ACM.
Davenport, T. H., & Prusak, L. (2000). Working knowledge: How organizations manage what they
Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of
Dinh, H. T., Lee, C., Niyato, D., & Wang, P. (2013). A survey of mobile cloud computing:
13(18), 1587-1611.
Field, A. (2005). Discovering statistics using SPSS. Thousand Oaks, CA: SAGE.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behavior: An introduction to theory
Hair, J. F., Tatham, R. L., Anderson, R. E., & Black, W. (2006). Multivariate data analysis (6th
Holte, R. C. (1993). Very simple classification rules perform well on most commonly used datasets.
Kim, S. H., & Kim, J. K. (2018). Determinants of the adoption of mobile cloud computing services:
Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). New York:
Kotsiantis, S. B., Zaharakis, I., & Pintelas, P. (2007). Supervised machine learning: A review of
160, 3-24.
Lee, Y. C. (2016). Why do people adopt cloud services? Gender differences. Social Science
Leech, N. L., Barrett, K. C., & Morgan, G. A. (2005). SPSS for intermediate statistics: Use and
Lian, J. W., Yen, D. C., & Wang, Y. T. (2014). An exploratory study to understand the critical
factors affecting the decision to adopt cloud computing in Taiwan hospital. International Journal
does it also add value? In Knowledge Management: Current Issues and Challenges (pp. 66-78).
Ngai, E. W., Xiu, L., & Chau, D. C. (2009). Application of data mining techniques in customer
relationship management: A literature review and classification. Expert Systems with Applications,
36(2), 2592-2602.
Nistor, N. (2014). When technology acceptance models won’t work: Non-significant intention-
5(1), 14-37.
Nonaka, I., & Takeuchi, H. (1995). The knowledge-creating company: How Japanese companies
Park, E., & Kim, K. J. (2014). An integrated adoption model of mobile cloud services: Exploration
of key determinants and extension of technology acceptance model. Telematics and Informatics,
31(3), 376-385.
Rogers, E. M. (2003). Diffusion of innovations (5th ed.). New York, NY: Free Press.
Sharma, S. K., Al-Badi, A. H., Govindaluri, S. M., & Al-Kharusi, M. H. (2016). Predicting
Sharma, S. K., Joshi, A., & Sharma, H. (2016). A multi-analytical approach to predict the Facebook
Shiau, W. L., & Chau, P. Y. (2016). Understanding behavioral intention to use a cloud computing
classroom: A multiple model comparison approach. Information & Management, 53(3), 355-365.
Shon, T., Cho, J., Han, K., & Choi, H. (2014). Toward advanced mobile cloud computing for the
internet of things: current issues and future direction. Mobile Networks and Applications, 19(3),
404-413.
Stergiou, C., & Psannis, K. E. (2017). Recent advances delivered by mobile cloud computing and
internet of things for big data applications: A survey. International Journal of Network
Subashini, S., & Kavitha, V. (2011). A survey on security issues in service delivery models of
Tan, G. W. H., Ooi, K. B., Leong, L. Y., & Lin, B. (2014). Predicting the drivers of behavioral
intention to use mobile learning: A hybrid SEM-Neural Networks approach. Computers in Human
Turban, E., Sharda, R., & Delen, D. (2011). Decision support and business intelligence systems.
Ul Amin, R., Inayat, I., Shahzad, B., Saleem, K., & Aijun, L. (2017). An empirical study on
acceptance of secure healthcare service in Malaysia, Pakistan, and Saudi Arabia: a mobile cloud
Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of information
Wang, C. S., & Huang, Y. M. (2016). Acceptance of cloud services in face-to-face computer-
supported collaborative learning: A comparison between single-user mode and multi-user mode.
explained
Behavioral CUI1. “I intend to continue to use mobile .91 .76 .85 .72 74.34 .91 .68
future.”
purposes.”
24
information management.”
Attitude A1. “Using mobile cloud services for .89 .73 .82 .67 65.30 .89 .59
pleasant.”
idea.”
25
interesting.”
A6. “I like the idea of using mobile cloud .67 .77 .60
management.”
Informatio IR1. “Using mobile Internet on my .91 .75 .86 .74 78.43 .91 .51
anywhere.”
my smartphone.”
26
Informatio IS1. “Using mobile cloud services on my .91 .81 .91 .83 84.06 .91 .52
smartphone.”
Informatio ISh1. “Using social media applications on my .71 .51 .80 .64 64.61 .71 .51
Informatio IA1: “I apply knowledge and experience .83 .64 .83 .69 74.72 .83 .53
processes.”
28
problem‐solving activities.”
ACCEPTED MANUSCRIPT
29
Note: A= Attitude; BI= Behavioral Intention; IR= Information Retrieval; ISt= Information Storage;
10-fold .648 .089 .687 .648 .667 .571 .780 .526 Low
cross- .845 .430 .770 .845 .806 .434 .708 .748 Medium
validation .395 .042 .607 .395 .479 .427 .677 .324 High
Percentage .650 .118 .565 .650 .605 .505 .766 .434 Low
split .800 .429 .789 .800 .794 .374 .686 .764 Medium
(66%) .400 .056 .545 .400 .462 .394 .672 .304 High
Naive Bayes 67.86 64.76 .679 .648 .664 .647 .666 .642
1CCI: Correctly Classified Instances, 2TP: True Positive, 3PS: Percentage Split
Information
Retrieval
Information
Storage
.77**
Behavioral
Attitude
Intention
Information R2 = .33, e = .17 R2 = .59, e = .17
Sharing
Information
Application
Educational use of the MCC services in the higher education was investigated.
A hybrid modeling approach combining both machine learning and SEM was employed.
The structural model explained 59% of the variance in the behavioral intentions.
Students’ information management practices were associated with the attitudes.
OneR algorithm predicted the behavioral intentions with an accuracy of 74%.